Seminar Dr Pierre Larmande
AI for Plant Regulatory Genomics: DNA Language Models, Graph Neural Networks, and FAIR Knowledge Graphs - Tuesday, January 13th 2025 11 am, INRAE, Versailles
Understanding how genomic information drives plant development, stress responses, and phenotypic diversity requires the ability to interpret regulatory sequences, integrate multi-omics datasets, and connect molecular mechanisms across scales. My research contributes to this challenge by developing computational approaches that combine FAIR knowledge infrastructures with modern AI methods to advance regulatory genomics in plants.
Building on AgroLD, a large FAIR Knowledge Graph that integrates hundreds of plant genomics and phenomics resources, we investigate how structured biological knowledge can be combined with sequence-based models to better understand gene regulation. Recent work includes the evaluation of DNA Language Models (LLMs) and deep learning approaches for characterizing regulatory regions in plant genomes, and the extension of these approaches toward predictive modeling of regulatory behavior. Our current efforts explore multimodal AI architectures that couple sequence-level representations with graph-based reasoning, enabling more comprehensive analyses of regulatory processes in plant genomes.
Together, these contributions aim to deepen our understanding of how regulatory information is encoded in plant genomes, support more accurate interpretation of gene function and stress responses, and provide new computational frameworks to accelerate discovery in plant biology and crop improvement.
Pierre Larmande is a senior research scientist at the French Institute of Research for Sustainable Development - IRD, based in Montpellier, in the "Cereal root systems" CERES team, DIADE unit. His research focuses on bioinformatics for plant sciences, with expertise in knowledge graphs, semantic web technologies, data integration, and AI-driven knowledge discovery.
Invitation: Sylvie Coursol
"Biomass Quality and Interactions with Drought" QUALIBIOSEC team
In connection with the research developed at the Institute Jean-Pierre Bourgin for Plant Sciences.
Building on AgroLD, a large FAIR Knowledge Graph that integrates hundreds of plant genomics and phenomics resources, we investigate how structured biological knowledge can be combined with sequence-based models to better understand gene regulation. Recent work includes the evaluation of DNA Language Models (LLMs) and deep learning approaches for characterizing regulatory regions in plant genomes, and the extension of these approaches toward predictive modeling of regulatory behavior. Our current efforts explore multimodal AI architectures that couple sequence-level representations with graph-based reasoning, enabling more comprehensive analyses of regulatory processes in plant genomes.
Together, these contributions aim to deepen our understanding of how regulatory information is encoded in plant genomes, support more accurate interpretation of gene function and stress responses, and provide new computational frameworks to accelerate discovery in plant biology and crop improvement.
Pierre Larmande is a senior research scientist at the French Institute of Research for Sustainable Development - IRD, based in Montpellier, in the "Cereal root systems" CERES team, DIADE unit. His research focuses on bioinformatics for plant sciences, with expertise in knowledge graphs, semantic web technologies, data integration, and AI-driven knowledge discovery.
Invitation: Sylvie Coursol
"Biomass Quality and Interactions with Drought" QUALIBIOSEC team
In connection with the research developed at the Institute Jean-Pierre Bourgin for Plant Sciences.
Back
